From R¶

The c4r package loads data from R through the same data catalog as Python's cohort(), and returns ordinary data.frames. It is a thin wrapper: the Python c4 package does the work, through reticulate.

Install¶

You need the Python install first (setup), plus the R package reticulate. Then, from the root of the c4 repository:

R CMD INSTALL c4r

or, in R, install.packages("c4r", repos = NULL, type = "source").

Set up¶

Tell c4r which Python has c4 installed, once per session:

library(c4r)
c4r_setup(python = "/path/to/venv/bin/python")   # or a virtualenv directory

To skip this step, set C4_PYTHON in ~/.Renviron:

C4_PYTHON=/path/to/venv/bin/python

To use another catalog, e.g. the demo one, pass it to setup: c4r_setup(catalog = "examples/demo/catalog.yaml").

Check the setup

From the repository root, C4_PYTHON=/path/to/venv/bin/python Rscript c4r/tests/run_demo.R runs every method on the demo cohort and ends with "All c4r checks passed."

Load data¶

The cohort object has the same methods as in Python, called with $ instead of .. The examples use the fake demo cohort, shown with its data folder as /data/lab.

c4_cohorts()

[1] "demo"

demo <- cohort("demo")          # prints what's available, as in Python
pheno <- demo$load("pheno")
str(pheno)

Left out 1 patient(s) marked drop=1 in the id_map (keep_dropped=True to keep them).
'data.frame':   7 obs. of  8 variables:
 $ patient_id: chr  "P001" "P002" "P003" "P004" ...
 $ qc_note   : chr  NA NA "old exome_id: EX-0003" "old exome_id: EX-0004" ...
 $ ID        : chr  "P001" "P002" "P003" "P004" ...
 $ age       : int  64 58 71 66 49 73 55
 $ female    : int  0 1 1 0 1 0 0
 $ site      : chr  "A" "A" "B" "B" ...
 $ OS_mo     : num  31.2 44 12.5 27.9 50.1 19 38.6
 $ OS_event  : int  1 0 1 1 0 1 0

Several table names are merged on patient_id:

df <- demo$load("pheno", "pcs", "hla.hlahd")
df[, c("patient_id", "OS_mo", "IID", "PC1", "seqID", "A")]

Note: 1 of 7 rows in 'pcs' have no patient_id (array ID not in id_map).
Left out 1 patient(s) marked drop=1 in the id_map (keep_dropped=True to keep them).
  pheno.main                      7 patients
  (1 row(s) of pcs without a patient_id dropped before merging)
+ pcs                             5 matched   (2 without)
+ hla.hlahd.standardized          5 matched   (2 without)
= 7 rows (left join on patient_id)
  patient_id OS_mo      IID    PC1   seqID               A
1       P001  31.2 ARR-0001  0.012 EX-0001 A*02:01/A*24:02
2       P002  44.0 ARR-0002  0.010 EX-0002 A*01:01/A*03:01
3       P003  12.5 ARR-0003 -0.044 EX-0004 A*11:01/A*24:02
4       P004  27.9 ARR-0004  0.009 EX-0003 A*02:01/A*02:01
5       P005  50.1     <NA>     NA EX-0005 A*03:01/A*68:01
6       P007  19.0 ARR-0007 -0.051    <NA>            <NA>
7       P008  38.6     <NA>     NA    <NA>            <NA>

$load() arguments - Unnamed: table names. - Named: values for templated files, e.g. demo$load("hla.hibag", panel = "african"). - Options: how ("left", "inner", "outer"), qc_note, keep_dropped, verbose.

What it does is described in Loading and merging.

Everything else¶

demo$path("dna.imputed", chrom = 6)        # "/data/lab/demo/dna/chr6.dose.vcf.gz"
demo$paths("dna.imputed")[["6"]]           # all chromosomes, named by number
demo$describe("OS_mo")                     # Overall survival (months) (numeric)
demo$id_map()                              # patient_id / exome_id / array_id / imputed_id
demo$dropped_patients()                    # "P006"
demo$check()                               # problems with the catalog, if any

How R and Python differ¶

	Python	R
Call a method	`demo.load("pheno")`	`demo$load("pheno")`
Options	`keep_dropped=True`	`keep_dropped = TRUE`
Missing values	`NaN` / `None`	`NA`
HLA genotype columns	`HLAgenotype` objects	text, e.g. `"A02:01/A24:02"`
Merge report and notes	printed	R messages; hide with `verbose = FALSE` or `suppressMessages()`
Errors	Python exceptions	R errors with the same message

For example, an invalid template value:

demo$load("hla.hibag", panel = "martian")

Error: ValueError: Invalid panel='martian' for 'hla.hibag.raw'. Choose from: ['european', 'african', 'asian', 'multiethnic']

Only data loading is wrapped. The command-line tools (c4 dosages, c4 survival, …) work the same whether you use R or Python; see Command line.